Tube DIY Asylum

Do It Yourself (DIY) paradise for tube and SET project builders.

Return to Tube DIY Asylum

Message Sort: Post Order or Asylum Reverse Threaded

musical loadlines

67.189.165.161
Posted on July 20, 2015 at 09:54:35
dave slagle
Manufacturer

Posts: 5430
Location: NYC
Joined: April 27, 2001
 This is an offshoot from a thread on SET... feel this is a more appropriate venue.

The general question is "how do you size the inductance for loading a triode"

The two sides are:

The inductance value required depends on the reflected load

The inductance value required depends on the source impedance (Triode Rp)

I fall into the source Z determining the inductance (and may be the only one) and in order to start things off I'll add a few pictures.

The signal being traced starts with a couple of full scale 20hz sines followed by the cannon fires from the 1812 and then closes with a couple of 20hz sines.



When looked at as a loadline on some 2A3 curves we get the following. 2A3 is operating at book points with a 25hy inductive load. The green represents a 2K5 reflected load and the blue a 10K reflected load.



dave


 

Hide full thread outline!
    ...
Very interesting Concept!, posted on July 20, 2015 at 13:32:42
gusser
Audiophile

Posts: 3649
Location: So. California
Joined: September 6, 2006
 That is using .wav or other audio files with SPICE.

Should be applicable with SS amp too.

 

Got Awfully Hungry !!, posted on July 20, 2015 at 14:51:11
drlowmu
Manufacturer

Posts: 9730
Location: East of Kansas City
Joined: January 10, 2005
 After carefully studying your screen shots, I decided what it looked like "to me". Then, I immediately sought the best Italian Restaurant in my college town, for their biggest plate of Spaghetti !! It was delicious, BTW, as IS your screen shot and your fine mind, dear Dave !!

Jeff Medwin

 

okay, so what does it tell us? and how is it relevant to a classical theory PP rebuild?, posted on July 20, 2015 at 16:28:16
Timbo in Oz
Audiophile

Posts: 23221
Location: Canberra - in the ACT - SE Australia
Joined: January 30, 2002
 like mine?

TIA Dave.


Warmest

Tim Bailey

Skeptical Measurer & Audio Scrounger


 

RE: musical loadlines, posted on July 20, 2015 at 16:46:19
Paul Joppa
Industry Professional

Posts: 7295
Location: Seattle, WA
Joined: April 23, 2001
 The "book [operating] point" for a 2A3 is optimized for a 2500Hz load, so reflecting a 10K load is unrealistic. If the operation were optimized for a 10K load (425v at 35mA) clipping would obviously take place on the 20Hz sinewaves. That's why, for large signal amplitude, you need an inductance tied to the load impedance.

 

5X the Operating Z, posted on July 20, 2015 at 18:08:11
Triode_Kingdom
Audiophile

Posts: 10042
Location: Central Texas
Joined: September 24, 2006
 "The inductance value required depends on the reflected load"

The reflected load is the operating impedance of the circuit. Rule of thumb for acceptable loss resulting from a shunt inductor is the XL should be no less than 5X the operating impedance. More is better, all else being equal.




 

RE: 5X the Operating Z, posted on July 20, 2015 at 18:38:03
dave slagle
Manufacturer

Posts: 5430
Location: NYC
Joined: April 27, 2001
 I'll agree that in a perfect world... more is better but in the real world, what is the "acceptable loss" caused by finite inductance?

I think there are two extremes at play here. One extreme is a plate choke attempting to deliver maximum voltage with very little current (large value load) and the other an output transformer attempting to deliver maximum current. What I am interested in is discussing the huge grey area inbetween.

When it comes to current delivery to a speaker through an output transformer, increasing the load value increases distortion since the increase of turns ratio requires more voltage swing and any gains from making the loadline more horizontal are lost by the increased voltage demand. This follows the same lines of for a given power output, an 845 will have less distortion than a 45.

dave

 

Seems to follow accepted theory to me ..., posted on July 21, 2015 at 05:10:08
Naz
Audiophile

Posts: 2184
Location: Sydney
Joined: September 2, 2005
 At 20Hz the loadline is more vertical as expected, caused by the lower than ideal inductance. At the higher frequencies of the audio signal the loadline is more ideal but still elliptical caused by the limitations of the transformer (leakage C & L) at high and low frequency extremes.

I was more interested in what happened in your earlier example where the music was played simultaneously with the LF tones. As expected the music rode on the LF waveform which could aid to premature clipping as Paul pointed out.

However I believe that at any single point along the non ideal LF loadline the more ideal loadline for the higher audio frequencies is preserved. Thus lower distortion at mid frequencies as opposed to frequencies at the extremes.

The width of the the loadline at LF is concerning but perhaps just serves to highlight the limitations of transformers?

Naz


 

Ahh, totally agree with this statement ...., posted on July 21, 2015 at 05:16:29
Naz
Audiophile

Posts: 2184
Location: Sydney
Joined: September 2, 2005
 "This follows the same lines of for a given power output, an 845 will have less distortion than a 45"

For the reasons you explained this has been my belief ... well it's more than a belief, it's predictable and measurable.

Naz


 

RE: Seems to follow accepted theory to me ..., posted on July 21, 2015 at 07:49:52
Tre'
Industry Professional

Posts: 17293
Location: So. Cal.
Joined: February 9, 2002
 In the presents of both HF and LF.......when a large HF signal is at the lower part of the large ellipse, caused by the LF, the HF will run out of current early.

The only way to prevent this is to either increase the inductance or limit the low frequencies.

Tre'


Have Fun and Enjoy the Music
"Still Working the Problem"

 

RE: Seems to follow accepted theory to me ..., posted on July 21, 2015 at 11:47:44
dave slagle
Manufacturer

Posts: 5430
Location: NYC
Joined: April 27, 2001
 In the presents of both HF and LF.......when a large HF signal is at the lower part of the large ellipse, caused by the LF, the HF will run out of current early.

I disagree. Everything is held constant in this but the load where green is 2K5 and blue is 10K.



It seems clear to me that the situation where current is going to become the issue is with the heavier load. Now lets look at the case of a plate choke. Green is 2K5 load and blue is 1 meg.





Now if the "width of the loop" is really a major concern, one would expect a plate choke to be far less than optimal and a CCS to be the ideal, yet I have yet to hear a CCS better a plate choke. The natural response to this from the "ccs camp" is I simply must enjoy distortion. For me, as long as the lines are closer to horizontal and keep away from the current extremes, I don't lose much sleep over it. I could really go out on a limb and say that for a given voltage swing, the larger the load, the more inductance you need to avoid traversing the nonlinear parts of the curves.

dave

 

RE: musical loadlines, posted on July 22, 2015 at 02:11:51
shane.
Audiophile

Posts: 394
Location: S.A.
Joined: March 27, 2004
 WRT original question, I'd answer XL = 2x Rp at lowest usable (system) frequency.

WRT to OP, did you try Gary Pimms CCS?

Shane

 

RE: Seems to follow accepted theory to me ..., posted on July 22, 2015 at 19:56:42
Tre'
Industry Professional

Posts: 17293
Location: So. Cal.
Joined: February 9, 2002



Just do a sim with 20Hz and 1000Hz at the same time with both being large signals.

Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

RE: Seems to follow accepted theory to me ..., posted on July 23, 2015 at 06:27:25
dave slagle
Manufacturer

Posts: 5430
Location: NYC
Joined: April 27, 2001



OK... well they can't both be "large signals" since they will clip if the values sum to more than the bias voltage so I did 22.5V+22.5V. The above was for 5 "loops" and the width of the loop isn't as wide as it appears since it takes a cycle or two to "settle" at the bias point. Below is a single loop and the actual loop width is about 2/3 of what shows up.





 

RE: Seems to follow accepted theory to me ..., posted on July 23, 2015 at 10:33:04
Tre'
Industry Professional

Posts: 17293
Location: So. Cal.
Joined: February 9, 2002

" well they can't both be "large signals" since they will clip if the values sum to more than the bias voltage so I did 22.5V+22.5V. "

I see what you mean.

But isn't there still a problem with the 1kHz operating at different points as it follows the ellipse? When it's at the lower right vs the upper left?

Won't that lead to some form of IMD?

What does the above sim look like mixing 1kHz with 200Hz instead of 20Hz?

Thanks.

Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

RE: Seems to follow accepted theory to me ..., posted on July 29, 2015 at 09:33:56
dave slagle
Manufacturer

Posts: 5430
Location: NYC
Joined: April 27, 2001
But isn't there still a problem with the 1kHz operating at different points as it follows the ellipse? When it's at the lower right vs the upper left?

If the lines were all parallel and evenly spaced I would say no but as you get to the knee, i would expect distortion.

Won't that lead to some form of IMD?

yes and i would expect the distortion of the 2K5 load to be worse since it traverses a more nonlinear part of the curves.

What does the above sim look like mixing 1kHz with 200Hz instead of 20Hz?





for anyone still interested here is another set of plots where i changed the inductance and the load.





I guess the point I am trying to get to is the actual width of the ellipse may not be as important as its slope.

dave



 

RE: Seems to follow accepted theory to me ..., posted on July 29, 2015 at 14:02:37
Tre'
Industry Professional

Posts: 17293
Location: So. Cal.
Joined: February 9, 2002

Thank you Dave.

Keeping the low frequency out of the picture keeps the ellipse from being so large (wide) that it drives the signal into the more extremely non-linear parts of the tube's curves, as shown in the green and red of your second graph.

In the end, I think that was all I was trying to say. I also think that is what Voltsecond was saying.

All I know for sure is when I limited the LF going to my SET amp (I already have woofers, driven by a SS amp, handling the bass and my mids don't play bass anyway) the sound of the mids and highs got much better.

I believe the graph above, vs. the other graph you show, is the reason for this.

Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

RE: Seems to follow accepted theory to me ..., posted on July 29, 2015 at 21:07:07
dave slagle
Manufacturer

Posts: 5430
Location: NYC
Joined: April 27, 2001

I guess then the questions become....How do you filter? and is the cure worse than the sickness?

The image above represents a nearly full scale 20hz signal which seldom happens in real music and if it does happen it is not cyclical for any period of time. Below is a 10 second section of the 1812 with the cannon fires. 25hy 2K5 and 10K loads.





The question that needs to still be answered is which is worse, the narrower green envelope that spans 110ma of the curves or the wider blue envelope that spans 55ma? To me the blue remains in the more linear part of the curves and keeps with my building / listening experience. FWIW, the flat line in the lower right of the curves is hard clipping of the .wav file.

dave

 

RE: Seems to follow accepted theory to me ..., posted on July 30, 2015 at 08:38:48
Tre'
Industry Professional

Posts: 17293
Location: So. Cal.
Joined: February 9, 2002
 "I guess then the questions become....How do you filter? and is the cure worse than the sickness?"

That is clearly up for debate.

Not just the single series filter cap but the fact that's the rest of the frequencies are being played by a SS amp.

Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

RE: Seems to follow accepted theory to me ..., posted on August 12, 2015 at 08:21:31
Mr_Steady
Audiophile

Posts: 2042
Location: North Florida
Joined: August 19, 2014
 It would be interesting to me to see what sim would look like for a midrange amp, say 400hz and 6000hz.




---------------------------------------------------
Big speakers and little amps blew my mind!

 

Page processed in 0.037 seconds.

[ Top Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2024, Audio Asylum, all rights reserved ]
[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]